GlycoSuiteDB, a database of glycan structures, has been constructed with an emphasis on quality, consistency and data integrity. Importance has been placed on making the database a reliable and useful resource for all researchers. This database can help researchers to identify what glycan structures are known to be attached to certain glycoproteins, as well as more generally identifying what types of glycan structures are associated with different states, for example, different species, tissues and diseases. To achieve this, a major effort has gone into data standardisation. Many rules and standards have been adopted, especially for representing glycan structure and biological source information. This paper describes some of the challenges faced during the continuous development of GlycoSuiteDB.

Visualization interfaces for high performance computing systems pose special problems due to the complexity and volume of data these systems manipulate. In the post-genomic era, scientists must be able to quickly gain insight into structure-function problems, and require flexible computing environments to quickly create interfaces that link the relevant tools. Feature, a program for analyzing protein sites, takes a set of 3-dimensional structures and creates statistical models of sites of structural or functional significance. Until now, Feature has provided no support for visualization, which can make understanding its results difficult. We have developed an extension to the molecular visualization program Chimera that integrates Feature's statistical models and site predictions with 3-dimensional structures viewed in Chimera. We call this extension ViewFeature, and it is designed to help users understand the structural Features that define a site of interest. We applied ViewFeature in an analysis of the enolase superfamily; a functionally distinct class of proteins that share a common fold, the alpha/beta barrel, in order to gain a more complete understanding of the conserved physical properties of this superfamily. In particular, we wanted to define the structural determinants that distinguish the enolase superfamily active site scaffold from other alpha/beta barrel superfamilies and particularly from other metal-binding alpha/beta barrel proteins. Through the use of ViewFeature, we have found that the C-terminal domain of the enolase superfamily does not differ at the scaffold level from metal-binding alpha/beta barrels. We are, however, able to differentiate between the metal-binding sites of alpha/beta barrels and those of other metal-binding proteins. We describe the overall architectural Features of enolases in a radius of 10 Angstroms around the active site.

In combinatorial library design and use, the conformation space of molecules can be represented using three-dimensional (3-D) pharmacophores. For large libraries of flexible molecules, the calculation of these 3-D pharmacophoric fingerprints can require examination of trillions of pharmacophores, presenting a significant practical challenge. Here we describe the mapping of this problem to the UCSC Kestrel parallel processor, a single-instruction multiple-data (SIMD) processor. Data parallelism is achieved by simultaneous processing of multiple conformations and by careful representation of the fingerprint structure in the array. The resulting application achieved a 35+ speedup over an SGI 2000 processor on the prototype Kestrel board.

Significant progress in NMR methodology for measuring spin-relaxation data at many different 15N and 13C sites in proteins demands new and increasingly sophisticated ways of data interpretation. Recent work of our group concerning the use of anisotropic and reorientational collective motional models for spin-relaxation interpretation is briefly reviewed and a number of important aspects of collective reorientational motional models are discussed at the example of a 11 ns molecular dynamics computer simulation of the protein ubiquitin.

This work describes ANN-Spec, a machine learning algorithm and its application to discovering un-gapped patterns in DNA sequence. The approach makes use of an Artificial Neural Network and a Gibbs sampling method to define the Specificity of a DNA-binding protein. ANN-Spec searches for the parameters of a simple network (or weight matrix) that will maximize the specificity for binding sequences of a positive set compared to a background sequence set. Binding sites in the positive data set are found with the resulting weight matrix and these sites are then used to define a local multiple sequence alignment. Training complexity is O(lN) where l is the width of the pattern and N is the size of the positive training data. A quantitative comparison of ANN-Spec and a few related programs is presented. The comparison shows that ANN-Spec finds patterns of higher specificity when training with a background data set. The program and documentation are available from the authors for UNIX systems.

The information as to where and when a mRNA is present in a given cell is essential to bridge the gap between the DNA sequence of a gene and its physiological function. Therefore, a major component of functional genomics is to characterize the levels and the spatio-temporal domains of gene expression. Currently, there is just a few specialised public databases available storing the data on gene expression while they are needed as a resource for the field. Moreover, there is a need to develop and assess computational tools to compare and analyse expression profiles in a suitable way for biological interpretation. Here we describe our recent work on developing a database on gene expression for the frog Xenopus laevis, and on setting up and using new tools for the analysis and comparison of gene expression patterns. We used histogram clustering to compare expression profiles at both gene and tissue levels using a set of data coming from the characterization of the expression of genes during early development of Xenopus. This enabled us to draw a tree of tissue relatedness and to identify coexpressed genes by in silico analysis.

Selection-amplifications systems provide a means of engineering biomacromolecules with new properties. The combination of stringent functional selection with the ability to amplify single molecules confers great specificity on the evolving population. Yet such systems like many complicated chemical kinetic mechanisms can show a range of unstable and metastable behavior. These instabilities can be investigated using the Shannon entropy of the evolving population. It is shown that the Shannon entropy provides a Lyapounov function for exploring dynamic stability. A simple model of in vitro evolution is presented and stability conditions are established. It is seen that fairly simple directed evolution models can exhibit a range of dynamical behavior.

Information extraction technology, as defined and developed through the U.S. DARPA Message Understanding Conferences (MUCs), has proved successful at extracting information primarily from newswire texts and primarily in domains concerned with human activity. In this paper we consider the application of this technology to the extraction of information from scientific journal papers in the area of molecular biology. In particular, we describe how an information extraction system designed to participate in the MUC exercises has been modified for two bioinformatics applications: EMPathIE, concerned with enzyme and metabolic pathways; and PASTA, concerned with protein structure. Progress to date provides convincing grounds for believing that IE techniques will deliver novel and effective ways for scientists to make use of the core literature which defines their disciplines.

We present a new statistical approach for eukaryotic polymerase II promoter recognition. We apply stochastic segment models in which each state represents a functional part of the promoter. The segments are trained in an unsupervised way. We compare segment models with three and five states with our previous system which modeled the promoters as a whole, i.e. as a single state. Results on the classification of a representative collection of human and D. melanogaster promoter and non-promoter sequences show great improvements. The practical importance is demonstrated on the mining of large contiguous sequences.